Skip to contentSkip to conversation

kvm/powerpc/e500mc: fix tlb invalidation on cpu migration

PATCH In review Scott Wood 2 participants 3 messages Feb 19, 2013 – Apr 26, 2013

Subsystems (3)
lore
Share & tools
lore

Scott Wood

author
#0 lore raw

PATCH 1/1 kvm/powerpc/e500mc: fix tlb invalidation on cpu migration

The existing check handles the case where we've migrated to a different core than we last ran on, but it doesn't handle the case where we're still on the same cpu we last ran on, but some other vcpu has run on this cpu in the meantime.

---
This seems to have been the cause of the userspace segfaults I was seeing (the other TLB patches I posted are still needed as well).

 arch/powerpc/kvm/e500mc.c |    6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)
arch/powerpc/kvm/e500mc.c +5 -1
index 1f89d26..8637689 100644
@@ -111,6 +111,7 @@ void kvmppc_mmu_msr_notify(struct kvm_vcpu *vcpu, u32 old_msr)
 void kvmppc_core_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
 {
 	struct kvmppc_vcpu_e500 *vcpu_e500 = to_e500(vcpu);
+	static struct kvm_vcpu *last_vcpu_on_cpu[NR_CPUS];
 
 	kvmppc_booke_vcpu_load(vcpu, cpu);
 
@@ -136,8 +137,11 @@ void kvmppc_core_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
 	mtspr(SPRN_GDEAR, vcpu->arch.shared->dar);
 	mtspr(SPRN_GESR, vcpu->arch.shared->esr);
 
-	if (vcpu->arch.oldpir != mfspr(SPRN_PIR))
+	if (vcpu->arch.oldpir != mfspr(SPRN_PIR) ||
+	    last_vcpu_on_cpu[smp_processor_id()] != vcpu) {
 		kvmppc_e500_tlbil_all(vcpu_e500);
+		last_vcpu_on_cpu[smp_processor_id()] = vcpu;
+	}
 
 	kvmppc_load_guest_fp(vcpu);
 }
Signature · 1 line
1.7.9.5
#1 lore raw
Quoted 18 lines · Scott Wood

The existing check handles the case where we've migrated to a different core than we last ran on, but it doesn't handle the case where we're still on the same cpu we last ran on, but some other vcpu has run on this cpu in the meantime.

---
This seems to have been the cause of the userspace segfaults I was seeing (the other TLB patches I posted are still needed as well).

arch/powerpc/kvm/e500mc.c | 6 +++++-

1 file changed, 5 insertions(+), 1 deletion(-)
arch/powerpc/kvm/e500mc.c +0 -0
index 1f89d26..8637689 100644

arch/powerpc/kvm/e500mc.c void kvmppc_mmu_msr_notify(struct kvm_vcpu *vcpu, u32 old_msr) · line 114

Quoted 4 lines
arch/powerpc/kvm/e500mc.c +1 -0
@@ -111,6 +111,7 @@ void kvmppc_mmu_msr_notify(struct kvm_vcpu *vcpu, u32 old_msr)
 void kvmppc_core_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
 {
 	struct kvmppc_vcpu_e500 *vcpu_e500 = to_e500(vcpu);
+	static struct kvm_vcpu *last_vcpu_on_cpu[NR_CPUS];

Why not use DEFINE_PER_CPU?

Alex

Quoted 19 lines
arch/powerpc/kvm/e500mc.c +4 -1
 	kvmppc_booke_vcpu_load(vcpu, cpu);
 
@@ -136,8 +137,11 @@ void kvmppc_core_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
 	mtspr(SPRN_GDEAR, vcpu->arch.shared->dar);
 	mtspr(SPRN_GESR, vcpu->arch.shared->esr);
 
-	if (vcpu->arch.oldpir != mfspr(SPRN_PIR))
+	if (vcpu->arch.oldpir != mfspr(SPRN_PIR) ||
+	    last_vcpu_on_cpu[smp_processor_id()] != vcpu) {
 		kvmppc_e500_tlbil_all(vcpu_e500);
+		last_vcpu_on_cpu[smp_processor_id()] = vcpu;
+	}
 
 	kvmppc_load_guest_fp(vcpu);
 }
Signature · 1 line
1.7.9.5
#2 lore raw
In reply to Scott Wood
Quoted 13 lines · On 25.04.2013, at 11:04, Caraman Mihai Claudiu-B02008 wrote:
Quoted 13 lines · Scott Wood
Quoted 11 lines

The existing check handles the case where we've migrated to a different core than we last ran on, but it doesn't handle the case where we're still on the same cpu we last ran on, but some other vcpu has run on this cpu in the meantime.

Without this, guest segfaults (and other misbehavior) have been seen in smp guests.

Quoted 6 lines
Quoted 4 lines

Thanks, applied to kvm-ppc-3.9.

Alex

Can you pull it into kvm-ppc-queue?

It's not in kvm/next, so it'd potentially screw things up again when I create a new pull request.

Please just do a manual git merge of kvm/master or linus/master into kvm-ppc-queue.

Alex